The ability to assess the state of the fundus in patients with lens opacities of varying intensity, including patients with type 2 diabetes mellitus, using quantitative analysis of images made with a fundus camera. A pilot study

Purpose. To evaluate a quantitative method for grading the "blurriness" of ocular fundus images in patients with varying degrees of cataract including diabetes mellitus type II in order to improve the quality of fundus examination when it is affected by opacification of the crystalline lens, and to develop an objective method for measuring severity of cataract. Material and methods. We studied 16 patients (32 eyes) with a range of lens opacities and etiologies, including diabetes mellitus type II (12 patients). Age range was 70.19 ± 7.9 years. Degree of transparency of the lens was graded on the LOCS III system. Visual acuity was determined with 20/200 ETDRS at a distance of 4 m (ESV-3000). Images of the fundus were taken with a Topcon Mark II TRC retinal camera NW7SF Type IA. Image Resolution was 8 TIFF. An image processing algorithm that combined bandpass filtering, entropy analysis. and power spectral integration was used to quantify image haziness in terms of two inversely-related scales: “Blur Score” and “Clarity Score”. Data were analyzed via Spearman's rank correlation coefficient. Results. In order to better fix the details of the fundus images with lens opacities using a computer algorithm with a filter system, evaluation scales have been developed – the clarity score scale and the blur score scale. The two scales provide similar information, but in opposite directions. LOCS III lens opacities (NC and C) showed a positive correlation with Blur Score and negative correlation with Clarity Score. This indicates that a lower Blur Score, or higher Clarity Score, correlates with better visual acuity. The values of LOCS III NO and P, on the other hand, showed no reliable correlation with Blur or Clarity scores. Conclusions. A computer quantitative image analysis method originally developed to determine the opacities of the vitreous body can be used to analyze fundus images of patients with different stages of cataract when processing original images from a fundus camera. The proposed algorithm can be used to develop an objective method.